{"id":"https://openalex.org/W4407248372","doi":"https://doi.org/10.48550/arxiv.2502.04306","title":"ScoreFlow: Mastering LLM Agent Workflows via Score-based Preference Optimization","display_name":"ScoreFlow: Mastering LLM Agent Workflows via Score-based Preference Optimization","publication_year":2025,"publication_date":"2025-02-06","ids":{"openalex":"https://openalex.org/W4407248372","doi":"https://doi.org/10.48550/arxiv.2502.04306"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2502.04306","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2502.04306","pdf_url":"https://arxiv.org/pdf/2502.04306","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2502.04306","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5077043575","display_name":"Yinjie Wang","orcid":"https://orcid.org/0000-0001-7109-8736","observed_orcids":["https://orcid.org/0000-0001-7109-8736"]},"institutions":[{"id":"https://openalex.org/I40347166","display_name":"University of Chicago","ror":"https://ror.org/024mw5h28","country_code":"US","type":"education","lineage":["https://openalex.org/I40347166"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wang, Yinjie","raw_affiliation_strings":["University of Chicago"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Chicago","institution_ids":["https://openalex.org/I40347166"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113654217","display_name":"Ling Yang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I20089843","display_name":"Princeton University","ror":"https://ror.org/00hx57361","country_code":"US","type":"education","lineage":["https://openalex.org/I20089843"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yang, Ling","raw_affiliation_strings":["Princeton University "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Princeton University ","institution_ids":["https://openalex.org/I20089843"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100759678","display_name":"Guohao Li","orcid":"https://orcid.org/0000-0003-0260-5129","observed_orcids":["https://orcid.org/0000-0003-0260-5129","https://orcid.org/0009-0002-7072-1324","https://orcid.org/0009-0003-8057-7343"]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Li, Guohao","raw_affiliation_strings":["Project : https://github.com/ Gen-Verse/ScoreFlow","University of Oxford "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Project : https://github.com/ Gen-Verse/ScoreFlow","institution_ids":[]},{"raw_affiliation_string":"University of Oxford ","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100707461","display_name":"Mengdi Wang","orcid":"https://orcid.org/0000-0002-2207-2131","observed_orcids":["https://orcid.org/0000-0002-2207-2131"]},"institutions":[{"id":"https://openalex.org/I20089843","display_name":"Princeton University","ror":"https://ror.org/00hx57361","country_code":"US","type":"education","lineage":["https://openalex.org/I20089843"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wang, Mengdi","raw_affiliation_strings":["Princeton University "],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Princeton University ","institution_ids":["https://openalex.org/I20089843"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5047970850","display_name":"Bryon Aragam","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I40347166","display_name":"University of Chicago","ror":"https://ror.org/024mw5h28","country_code":"US","type":"education","lineage":["https://openalex.org/I40347166"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Aragam, Bryon","raw_affiliation_strings":["University of Chicago"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Chicago","institution_ids":["https://openalex.org/I40347166"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":80,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.2386000007390976,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.2386000007390976,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.16859999299049377,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.0966000035405159,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/llm-based-agents","display_name":"LLM agents","score":0.7990000247955322},{"id":"https://openalex.org/keywords/multi-agent-systems","display_name":"multi-agent systems","score":0.968999981880188},{"id":"https://openalex.org/keywords/preference-optimization","display_name":"preference optimization","score":0.7129999995231628},{"id":"https://openalex.org/keywords/gradient-based-optimization","display_name":"gradient-based optimization","score":0.7459999918937683},{"id":"https://openalex.org/keywords/continuous-optimization","display_name":"continuous optimization","score":0.6579999923706055},{"id":"https://openalex.org/keywords/question-answering","display_name":"question answering","score":0.8180000185966492},{"id":"https://openalex.org/keywords/coding","display_name":"coding","score":0.4950000047683716},{"id":"https://openalex.org/keywords/mathematical-reasoning","display_name":"mathematical reasoning","score":0.8790000081062317},{"id":"https://openalex.org/keywords/inference-cost-reduction","display_name":"inference cost reduction","score":0.4690000116825104}],"concepts":[{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.7149307727813721},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.664842963218689},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.46823424100875854},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3423864245414734},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.20616745948791504},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.1528218686580658},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.05482938885688782}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2502.04306","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2502.04306","pdf_url":"https://arxiv.org/pdf/2502.04306","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2502.04306","doi":"https://doi.org/10.48550/arxiv.2502.04306","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2502.04306","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2502.04306","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2502.04306","pdf_url":"https://arxiv.org/pdf/2502.04306","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4407248372.pdf","grobid_xml":"https://content.openalex.org/works/W4407248372.grobid-xml"},"referenced_works_count":29,"referenced_works":["https://openalex.org/W2736601468","https://openalex.org/W3177813494","https://openalex.org/W3195140297","https://openalex.org/W3210904070","https://openalex.org/W4327810158","https://openalex.org/W4377121468","https://openalex.org/W4384918448","https://openalex.org/W4386554921","https://openalex.org/W4387293261","https://openalex.org/W4389156617","https://openalex.org/W4390963096","https://openalex.org/W4392492277","https://openalex.org/W4393247779","https://openalex.org/W4399251554","https://openalex.org/W4399655920","https://openalex.org/W4401024643","https://openalex.org/W4402346081","https://openalex.org/W4403443836","https://openalex.org/W4403571374","https://openalex.org/W4403575202","https://openalex.org/W4403853635","https://openalex.org/W4405655184","https://openalex.org/W4405903187","https://openalex.org/W4407938709","https://openalex.org/W4409348251","https://openalex.org/W4415237471","https://openalex.org/W4415795501","https://openalex.org/W4415796092","https://openalex.org/W4416931028"],"related_works":["https://openalex.org/W4416373947","https://openalex.org/W7167904580","https://openalex.org/W6891781810","https://openalex.org/W4403571374","https://openalex.org/W4417051553","https://openalex.org/W4417289328","https://openalex.org/W4416037423","https://openalex.org/W7155000291","https://openalex.org/W7155789930","https://openalex.org/W7141535823"],"abstract_inverted_index":{"Recent":[0],"research":[1],"has":[2],"leveraged":[3],"large":[4],"language":[5],"model":[6],"multi-agent":[7],"systems":[8],"for":[9,88],"complex":[10],"problem-solving":[11],"while":[12],"trying":[13],"to":[14,20,38,114],"reduce":[15],"the":[16,24,81],"manual":[17],"effort":[18],"required":[19],"build":[21],"them,":[22],"driving":[23],"development":[25],"of":[26,43,80],"automated":[27],"agent":[28],"workflow":[29],"optimization":[30,52,69,84],"methods.":[31],"However,":[32],"existing":[33,107],"methods":[34],"remain":[35],"inflexible":[36],"due":[37],"representational":[39],"limitations,":[40],"a":[41,60,71,77],"lack":[42],"adaptability,":[44],"and":[45,98],"poor":[46],"scalability":[47],"when":[48],"relying":[49],"on":[50],"discrete":[51],"techniques.":[53],"We":[54],"address":[55],"these":[56],"challenges":[57],"with":[58,118],"ScoreFlow,":[59],"simple":[61],"yet":[62],"high-performance":[63],"framework":[64],"that":[65,86],"leverages":[66],"efficient":[67],"gradient-based":[68],"in":[70],"continuous":[72],"space.":[73],"ScoreFlow":[74,101],"incorporates":[75],"Score-DPO,":[76],"novel":[78],"variant":[79],"direct":[82],"preference":[83],"method":[85],"accounts":[87],"quantitative":[89],"feedback.":[90],"Across":[91],"six":[92],"benchmarks":[93],"spanning":[94],"question":[95],"answering,":[96],"coding,":[97],"mathematical":[99],"reasoning,":[100],"achieves":[102],"an":[103],"8.2%":[104],"improvement":[105],"over":[106],"baselines.":[108],"Moreover,":[109],"it":[110],"empowers":[111],"smaller":[112],"models":[113],"outperform":[115],"larger":[116],"ones":[117],"lower":[119],"inference":[120],"costs.":[121],"Project:":[122],"https://github.com/Gen-Verse/ScoreFlow":[123]},"counts_by_year":[{"year":2026,"cited_by_count":63},{"year":2025,"cited_by_count":17}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-02-08T00:00:00"}